Multidimensional Deprivation Spectrum: A Step Forward from Alkire

Abstract

Income-based poverty and multidimensional poverty are two major paradigms currently in use to define and measure poverty. Both these paradigms, however, take individuals as units of analysis and classify them on the basis of certain poverty lines and cut-offs as poor and non-poor. Social stigma and labelling theory suggest that the label of poverty negatively impacts the self-esteem of people or contributes to the tendencies of paternalistic dependency among them. This article suggests that poverty should be measured using dimensions of life as units of analysis. In this direction, it offers a variant of the Alkire and Foster (Counting and multidimensional poverty measures, OPHI Working Paper No. 7, Oxford Poverty & Human Development Initiative, 2007) multidimensional poverty index in the form of a multidimensional deprivation spectrum. Along with using different dimensions as units of analysis, the current article presents a whole spectrum of indices built to measure inequality for a more nuanced picture.

Keywords

Poverty measurement multidimensional deprivation poverty alleviation

Introduction

Since the advent of the construct of poverty it has been defined in terms of the lack of income (Chambers, 2006). This paradigm is based on the rationality assumption which argues for every human to be rational and use his income for taking care of his needs himself. However, Sen (1985) has found contrary evidence from his research on human life. Over time, the multidimensional nature of poverty was realized (Thorbecke, 2005). Income is a means to an end and not an end in itself, therefore for measuring poverty the direct measurement of all dimensions of human life such as education, health shelter, living standard, water and sanitation among others should be taken into consideration. Many efforts have been made so far as to capture this multidimensionality of poverty. The scope of this study, however, does not include the debate on the selection of dimensions for poverty measurement, for that is a different issue. This study focuses on an in-depth measurement of community poverty and its analysis.

The income-based poverty paradigm is the oldest and still the most commonly understood and used poverty definition around the globe. This paradigm has developed both in literature and operationalization over time. The World Bank uses this paradigm to measure poverty for international comparisons. The famous US$1 a day and US$2 a day definition of poverty is based on the income-poverty idea. It gave birth to many poverty measures for instance the head-count ratio, income-gap ratio, poverty-gap ratio, FGT measure (Foster, Greer, & Thorbecke, 1984) and many more. It has been in use for more than a century and it has grown to its apex. Nevertheless, it has not been free of criticism. The most prominent criticism is from the (Rawls, 1999) social justice theory which rejects income as the only measure of poverty of human life. Income and expenditure are not only subject to human preferences but also to external factors beyond human control. Income-based poverty measures fail to cover these external factors and the non-monetary deprivations (Bourguignon, 2002). A practical failure of this aspect of income-based measure is seen by recent improvement of developing countries in income poverty accompanied with a paradoxical increase in deprivations in many aspects of basic human needs and their redistribution (Besharov & Germanis, 2004; Citro & Micheal, 1995; Jencks, Mayer, & Swingle, 2004). Furthermore, income poverty requires direct transfer of money to the beneficiaries, which has its own challenges. These include true beneficiary selection, decision about amount of money transfer needed to alleviate poverty, training and awareness of beneficiaries in order to effectively use money allocated, monitoring and evaluation of such programmes to ensure success. However, even with the fulfilment of these conditions, the diversification of individual deprivations makes it impossible to achieve desirable results without health, education, empowerment and other dimensional provisions. The income paradigm fails to support such allocations. Finally, income measures are one-size-fits-all tools which do not cater for the differences in economic, social and other factors impacting incomes and expenditures.

In the light of the aforementioned shortfalls of the income paradigm, the need to develop an alternate definition and measure of poverty was felt, which resulted in the development of the multidimensional poverty paradigm. This alternate paradigm was based on Sen’s Capability Approach, which argues that deprivations should be measured on the basis of capabilities or endowments available to a person and not on the basis of choices he/she made. There have been practical advances by few researchers in this area. One of these measures was devised by Alkire and Foster (2007) and it had practical success by developing a multidimensional index of poverty capturing all different dimensions in the real sense. It has been used by the UNDP Human Development Report (UNDP, 2011) to access human deprivation globally. There is a lot of room for improvement for the multidimensional measures. These measures show a flexibility to adjust to the needs of individual countries which leads to a clearer picture of reality. Yet, these measures face limitations regarding, for instance, making decisions about the dimensions to be included, operationalization of these dimensions, complexity of measurement models and the generalization of the framework (Bourguignon, 2002).

Although when multidimensional poverty measures overcome the aforementioned limitations, it helps to depict a zoomed-in picture of poverty (Alkire & Seth, 2007) while focusing on individuals as units of analysis like the income measures. The classification of humans into categories of poor and non-poor appears to be the main goal of these measures. According to the social stigma theory and labelling theory the label of “poor” has an adverse effect on self-esteem of the people concerned and also develops a patronage dependency among them (Bos, Pryor, Reeder, & Stutterheim, 2013). The identification of people as “poor” is equivalent to remedial measures at the individual level and it will still require a selection of beneficiaries and a meeting of their requirements in terms of the dimensions of deprivation, which is an indirect way of measuring community deprivations.

The basic argument of the current study is that if a country knows its level of deprivation in different dimensions of education, health, living standard, etc., and there are direct provisions for attaining such endowments, then deprivations will be automatically eradicated at the individual level. For example, if poverty is high in the education dimension then the government will have to increase the educational facilities and budget. This will be helpful in two ways: first, it will save cost in terms of money and time to trace beneficiaries and second, it will help to create facilities which will help to develop the capabilities of communities.

The multidimensional deprivation spectrum (MDDS) is an extended form of the Alkire–Foster multidimensional poverty index. Identical to its parent index, it is a dual cut-off model where cut-offs are applied initially within dimensions to decide if an individual will be considered deprived. At the second stage, an overall cut-off is applied to detect the number of dimensions an individual has to be deprived off in order to be considered as poor or deprived overall. These are two major differences between spectrum deprivation analysis and the Alkire–Foster measure. The step forward in this analysis is of applying a spectrum of cut-offs at both the stages, resulting in the following classes: non-deprived borderline, slightly deprived, moderately deprived, highly deprived and absolutely deprived. Second, the multidimensional poverty spectrum does not locate individuals as poor or non-poor, it rather focuses on the total number of deprivations in each spectrum band in each dimension. This will help policymakers’ to recognize the exact extent and cause of deprivations on the basis of area, locality, gender, ethnic or any other class in order for them to design policy according to the specific targeted measures of deprivation alleviation. Similar to the Alkire–Foster methodology, an aggregation measure M₀ is calculated. It is a product of head count ratio H and average deprivation share A. However, here M is calculated twice, once at dimension level and then at the overall deprivation level an average M is calculated.

Methodology

Mathematically, it is a matrix approach where the rows represent the individuals and the columns represent dimensions. Let n represent the number of persons and N be the number of dimensions under consideration. Let y = [y_ij] denote the n × N matrix of achievements, where the typical entry y_ij > 0 is the achievement of individual i = 1, 2, …, n in dimension j = 1, 2, …, d. Each row vector y_i lists individual i’s achievements, while each column vector y_j gives the distribution of dimension j’s achievements across the set of individuals. In what follows we assume that d is fixed and given, while n is the sample size of the survey being used. Let z_0-5 > 0 denote the cut-offs below which a person is considered to be deprived at various spectrum levels in dimension j, and let z be the matrix of dimension specific cut-offs.

A methodology M for measuring multidimensional poverty consists of the identification method and aggregation method. The former is represented in such a way that ρ (y_j; z) = 5, if there is absolute deprivation in a particular dimension; ρ (y_j; z) = 4, if there is high deprivation in a particular dimension; ρ (y_j; z) = 3, if there is moderate deprivation in a particular dimension; ρ (y_j; z) = 2, if there is slight deprivation in a particular dimension; ρ (y_j; z) = 1, if it is a borderline case; and ρ (y_j; z) = 0, if there is no deprivation at all. Applying ρ (cut-off) to each dimension achievement vector in y yields the matrix Z {1, …, n} of dimension deprivations at various levels in y given z_0-5. The aggregation step then takes ρs as given for each level of spectrum and associates them with the matrix y and the cut-off matrix z for an overall level M (y; z) of multidimensional deprivation in each dimension. The resulting functional relationship M (adjusted headcount) is called an index, or measure of multidimensional deprivation.

M_{d} = M_{0} + M_{1} + M_{2} + M_{3} + M_{4} + M_{5}

M_{d} = (H_{0} \times A_{0}) + (H_{1} \times A_{1}) + (H_{2} \times A_{2}) + (H_{3} \times A_{3}) + (H_{4} \times A_{4}) + (H_{5} \times A_{5})

for each dimension.

Overall adjusted headcount M is then calculated as an average of dimension-wise adjusted headcounts,

M = \frac{M d 1 + M d 2 + M d 3 + \dots + M d N}{N} ​

where subscripts 0, 1, 2, 3, 4 and 5 represent not deprived, borderline case, slightly deprived, moderately deprived, highly deprived and absolutely deprived part of the spectrum, respectively. H = frequency of deprivation at each spectrum level/total sample size (n) and A = level of deprivation/maximum deprivation level.

Step-wise Approach

Step 1: Choose dimensions. For example, health, education, living standard, etc. (see Table 1).

Table 1.

Example for Application of Spectrum of Cut-offs at Dimensions Level (Step 1)

	Health			Education				Living Standard
	Immunization	Birthplace	Score	Read	Write	Math	Score	House	Electricity	Score
Person 1	5	1	6	1	5	0	6	5	0	5
Person 2	4	1	5	2	4	1	7	4	0	4
Person 3	3	1	4	3	3	2	8	3	0	3
Person 4	2	2	4	1	5	3	9	2	1	3
Person 5	1	1	2	2	4	4	10	1	2	3
Person 6	1	0	1	3	4	2	9	1	3	4

Source: The authors.

Step 2: Choose indicators. For example, immunization, birthplace for health or reading, writing, arithmetic skills and highest-class level for education, etc. (see Tables 2–4).

Table 2.

Score Range Application of First Identification Step

Score Range	Cut-off	Spectrum Band
0–1	0	ND
2	1	BLC
3	2	SD
4	3	MD
5	4	HD
6	5	AD

Source: The authors.

Note: Colours are only for visual aid of spectrum band differentiation.

Table 3.

Example for Application of Spectrum Band Cut-offs at Dimensions Level (Step 2)

	Health		Education		Living Standard
	Score	Spectrum Cut-off	Score	Spectrum Cut-off	Score	Spectrum Cut-off
Person 1	6	5	6	3	5	4
Person 2	5	4	7	4	4	3
Person 3	4	3	8	4	3	2
Person 4	4	3	9	5	3	2
Person 5	2	1	10	5	3	2
Person 6	1	0	9	5	4	3

Source: The authors.

Note: Colours are only for visual aid of spectrum band differentiation.

Step 3: Set poverty lines as per spectrum in two stages (for each indicator first apply AD, HD, MD, SD, BLC & ND and then do the scoring for each dimension). The first cut-offs step 1. (As this is a subjective measurement stage therefore, each of the dimensions will be graded on the basis of available responses.) In the current study as per participatory approach each participant defined his/her band score.

Step 4: In the second step of identification apply poverty lines. AD, HD, MD, SD, BLC & ND with the score range predefined for each value from 5 to 0 for the score of each dimension. For example, health spectrum cut-offs as shown in Table 2.

Similarly, for each dimension the scores should be translated into spectrum bands.

Step 6: Count the number of deprivations in each spectrum band. The per spectrum bands for health counts are shown in Table 4.

Table 4.

Count of Deprivations per Spectrum Band

Count	Spectrum Band
1	ND
1	BLC
0	SD
2	MD
1	HD
1	AD

Source: The authors.

Note: Colours are only for visual aid of spectrum band differentiation.

Demonstrating that out of six people one is not deprived, one is borderline case, two are moderately deprived, one is highly deprived and one is absolutely deprived.

Color coding for each band is done in Table 2 onwards for spectrum band differentiation.

Step 7: Calculate the Headcounts, H₀ – H₅ for each dimension. Divide the number of deprivations in each band with total sample size n.

Step 8: Calculate total score the deprivations in each dimension.

Step 9: Apply second cut-offs and redefine these score brackets into deprivation spectrum AD, HD, MD, SD, BLC and ND.

Step 10: Calculate the average deprivation gap (A) for each band. A is the average number of deprivations suffered in each spectrum band. It is calculated by adding up proportions of total deprivation score in a dimension.

Here if all six persons were absolutely deprived then the score would be 30. Therefore, the maximum score in each band (obtained by multiplying band cut-off with number of deprived people in the band) when divided by 30 results in average deprivation gap of the corresponding band of spectrum. This means a person who is deprived in no dimensions at all, is represented in the not deprived band must have 0 per cent weight in the analysis. This automatically removes the data of all non-deprived people from each dimension. When a person’s deprivation band changes, it changes the average deprivation gaps of both the bands depicting the change.

Step 11: Calculate the adjusted headcounts M for each dimension.

M_{d} = M_{0} + M_{1} + M_{2} + M_{3} + M_{4} + M_{5}

Which is calculated as a sum of the product of band-wise adjusted headcounts as follows:

M_{d} = (H_{0} \times A_{0}) + (H_{1} \times A_{1}) + (H_{2} \times A_{2}) + (H_{3} \times A_{3}) + (H_{4} \times A_{4}) + (H_{5} \times A_{5})

In the aforementioned example for the dimension of health it would be as follows:

\begin{matrix} M_{h} = (0.1667 \times 0) + (0.1667 \times 0.033) + (0 \times 0) + (0.3333 \times 0.2) + (0.1667 \times 0.133) + (0.1667 \times 0.08) \\ M_{h} = 0 + 0.0055 + 0 + 0.066 + 0.022 + 0.0136 \\ M_{h} = 0.107 \end{matrix}

This means the deprivation level in the dimension of health for this population is 10.7 per cent with 1.36 per cent absolute deprivation, 2.2 per cent high deprivation, 6.6 per cent moderate deprivation and 0.5 per cent borderline cases.

Step 12: Calculate the overall adjusted headcount M as an average of dimension-wise adjusted headcounts.

M = \frac{M d 1 + M d 2 + M d 3 + \dots + M d N}{N} ​

Empirical Testing

For empirically testing the MDDS methodology data was gathered from 1,799 female respondents from all the 36 districts of Punjab, Pakistan. A total of 50.1 per cent of the sample consisted of rural women and 49.9 per cent of urban females out of which 19.8 per cent were aged between 15 and 30 years, 24 per cent aged between 31 and 45 years, 35 per cent were between 46 and 60 years and 21.2 per cent were between 61 and 80 years. According to the work status approximately 30.4 per cent of total population consisted of females who did no work for the purpose of earning and they were labeled as housewives. A total of 40 per cent were menial workers or labourers, 23.6 per cent belonged to salaried class and 6 per cent of the sample was self-employed. Out of the total sample 29 per cent females were totally uneducated, 12.3 per cent attained primary education, 10.3 per cent were matriculate, 9.7 per cent were intermediate, 20.3 per cent had a bachelor’s degree, 17.8 per cent had a master’s degree and 0.4 per cent were postgraduates. The number of sons was also pointed out as a determinant of women empowerment by the participants during qualitative analysis, therefore this was added as an explanatory variable. A total of 27.8 per cent of the population was not married so this was not applicable to them; 28.7 per cent did not have any son at all; 32.6 per cent had 1 or 2 sons; 9.8 per cent of the sample had 3 or 4 sons; and only 1 per cent of the sample had 5 or 6 sons. When examined for the income-class 20.6 per cent of the sample was found to be marginalized, 40.5 per cent belonged to the middle-income group, 25.2 per cent considered themselves to be upper middle-income class and 13.7 per cent claimed to be rich. Table 5 represents the aforementioned findings.

Table 5.

Sample Demographics

		N	Marginal Percentage
Area	Rural	902	50.1
	Urban	897	49.9
Age (years)	15–30	357	19.8
	31–45	432	24.0
	46–60	629	35.0
	61–80	381	21.2
Work Status	House wife	547	30.4
	Menial/labour	719	40.0
	Salaried	425	23.6
	Self-employed	108	6.0
Education	Uneducated	522	29.0
	Primary	221	12.3
	Matric	186	10.3
	Intermediate	175	9.7
	Bachelors	366	20.3
	Masters	321	17.8
	Postgraduate	8	.4
Marital status	Unmarried	275	15.3
	Engaged	230	12.8
	Married	1,042	57.9
	Separated	47	2.6
	Divorced	52	2.9
	Widowed	153	8.5
Number of sons	NA	501	27.8
	None	517	28.7
	1 or 2	586	32.6
	3 or 4	177	9.8
	5 or 6	18	1.0
Income class	Marginalized	370	20.6
	Middle income	729	40.5
	Upper income	453	25.2
	Rich	247	13.7
Valid	1,799	100.0%
Missing	0
Total	1,799

Source: The authors.

Multidimensional Deprivation Spectrum

The multidimensional deprivation indices were created by applying a variant of the Alkire and Foster (2007) methodology. This method consists of 9 steps:

Step 1: In the first step, the unit of analysis is selected, which, in our case, was the household of the female participants.

Step 2: In the second step, the dimensions for analysis were selected. These dimensions were based on (Alkire & Foster, 2007) education, type of housing, electricity, assets, air quality, water, sanitation and income. The questions, definitions and poverty cut-off points of the dimensions and indicators used in steps 2 and 3 for the current study were developed on the basis of the demographic and health survey 2006–2007. The availability of international definitions of dimension and indicators along with the availability of national definitions of the same dimensions and indicators were taken into consideration. All the dimensions and indicators are defined on the basis of two categories for building a range of spectrum on a subjective basis, each dimension and indicator is divided into as many bands as there are to be defined. Here each dimension is in capital letters, followed by the indicators of that dimension in brackets. Questions for indicator measurement were used from Pakistan Demographics and Health Survey (PDHS) 2006–2007, which were as follows.

LIVING STANDARD: (type of house + electricity)

This dimension corresponds to Questions 109–111 and 107 in the PDHS questionnaire.

Question 109: Main material of floor (MFM): natural floor: earth/sand/mud floor; finished floor: chips/terrazzo, ceramic tiles, marble, cement, carpet, bricks, mats and others.

Question 110: Main material of roof (MRM): natural roofing: thatch/bamboo/wood/mud, rudimentary roofing, cardboard/plastic; finished roofing: iron sheets/asbestos, t-iron/wood/brick, reinforced brick cement/RCC and others.

Question 111: Main material of walls (MWM): natural walls: mud/stones, bamboo/sticks/mud; rudimentary walls: unbaked bricks/mud, plywood sheets, carton/plastic; finished walls: stone, blocks, baked bricks, cement blocks/cement, tent and others.

Question 107: Does your household (HH) has electricity? Yes/No

Poverty cut-off Z₁—In each question bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. Poverty cut-off denotes the situation under which a household is deprived in any two of the aforementioned indicators.

WATER & SANITATION: (source of drinking water + type of toilet facility)

This dimension corresponds to Questions 101 and 105 in the PDHS survey 2006–2007.

Question 101: Different sources of drinking water. Piped water: piped into dwelling, piped into yard, plot, public tap/stand pipe, tube well/borehole, hand pump; dug well: protected well, unprotected well; water from spring: protected spring, unprotected spring; rainwater; tanker truck; cart with small tank; surface water (river/dam/lake/pond/stream/canal); bottled water and others.

Question 105: Types of toilet facility. Flush or pour flush toilet: flush to sewer system, flush to septic tank, flush to somewhere else, flush to do not know where; pit latrine: ventilated improved, pit latrine with slab, pit latrine without slab, open pit; bucket toilet; hanging toilet/latrine; no facility/bush/field and others.

Poverty cut-off Z₂—In each question bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. The poverty cut-off denotes the situation under which a household is deprived in at least one of the aforementioned indicators.

AIR QUALITY: (type of cooking fuel)

This dimension corresponds to Question 108 in the PDHS survey 2006–2007.

Question 108: Types of cooking fuel in use: electricity, LPG, natural gas, biogas, kerosene, charcoal, wood, straw/shrubs/grass, agricultural crop, animal dung, no food cooked in house and others.

Poverty cut-off Z₃—In the above question bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. Poverty cut- off denotes the situation under which a household is deprived in at least any one of the aforementioned indicators.

ASSET: (access to different assets)

This dimension corresponds to Questions 107 and 114 in the PDHS survey 2006–2007.

Question 107: Does your HH have a refrigerator, TV, AC/room cooler and washing machine? Yes/No

Question 114: Does your HH have a car/truck? Yes/No

Poverty cut-off Z₄—If owns at least two out of these five assets. Bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. Poverty cut-off denotes the situation under which a household is deprived in at least any one of the above-mentioned indicators.

EDUCATION: (highest education level attained by the family members)

This dimension corresponds to Question V106 in the PDHS survey 2006–2007.

Question V106: Maximum education by any member.

Poverty cut-off Z₅—Maximum year of education completed by any member is less than 5 years. In each question bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. Poverty cut-off denotes the situation under which a household is deprived in at least any one of the aforementioned indicators.

LIVELIHOOD: (occupation of the respondent and her partner)

This dimension corresponds to Questions V705 and V717 in the PDHS survey 2006–2007.

Question V705: Respondent’s occupation: not working, professional/technical/managerial, clerical, sales, agriculture/self-employed, agricultural-employment, HH and domestic, services, skilled manual and unskilled manual.

Question V717: Partner’s occupation: not working, professional/technical/managerial, clerical, sales, agriculture/self-employed, agricultural-employment, HH and domestic, services, skilled manual and unskilled manual.

Poverty cut-off Z₆—In each question bold ones were considered as poor and allotted 1 value and non-bold ones were considered as non-poor and allotted 0 value. Poverty cut-off denotes the situation under which a household is deprived in at least any one of the aforementioned indicators.

Step 3: Third step consisted of choosing the indicators. Similarly, for the selection of indicators, the values assigned to each indicator band were chosen on the basis of questions and responses to the questions. This was commensurate to the subjective definitions of each band, similar to that of the example.

Step 4: This step involved the application of first cut-off of deprivation lines. In each indicator value 1 was assigned for all those options defined as deprived and value 0 was assigned for not the deprived options.

Step 5: This step consisted of the summation of total deprivation scores of each female. This variable was named as MDDS scores.

Step 6: In this step, a new variable MDDS was coded, when the total scores were redefined into the ranges of spectrum of deprivations. Any person with scores of 1–2 was recorded as 1 (not deprived), 3–4 as 2 (slightly deprived), 5–6 as 3 (moderately deprived), 7 as 4 (highly deprived) and 8 as 5 (absolutely deprived).

Step 7: This step consisted of calculation of average deprivation share “A” for each band of spectrum. To calculate A for each band/slab, the average deprivation per score included in the band is to be summed and the overall deprivation share is to be calculated by taking the average. For example, scores 1 and 2 are added to the first slab representing the not deprived band. For first spectrum band labeled as “not deprived” A₁ = (1/8 + 2/8)/2 = 0.1875. This represents that the individuals falling in the not deprived band have 18.75 per cent of deprivation. However, since it is a small percentage it has been labelled as not deprived.

Step 8: In this step the headcount H was calculated by using the frequency table generation for each band of the spectrum using the frequency table given below. For the not deprived band H₁ = 465/1799 = 0.258, it means 25.8 per cent females fall into the band of not deprived.

Step 9: The multidimensional deprivation spectrum is built using five indices, one for each band. Each index is the product of headcount and average deprivation share of each band. For example, for the not deprived band M₁ = A₁ × H₁ = 0.1875 × 0.258 = 0.048. This index is the adjusted head count ratio derived in the Alkire–Foster methodology (2007). It not only represents the incidence of deprivation in the not deprived band but, it also shows the depth of deprivation. The larger M represents both larger headcount and larger average deprivation share.

The rest of the indices of deprivation are summarized in Table 6. The table highlights the average share of deprivations per band and the percentage of deprived individuals per band, along with the index value.

Table 6 illustrates that females in the slightly deprived group faced 43.75 per cent deprivations of total possible deprivation and 26.8 per cent of the sample included females who were slightly deprived, with their index value being M₂ = 0.104. Similarly, those females who were labelled as moderately deprived faced 68.75 per cent of total possible deprivations, 15.2 per cent of the sample were moderately deprived and the highly deprived group which consists of 18.1 per cent of the sample were deprived in 87.5 per cent of the dimensions.

Table 6.

MDDS Band Scores

Band Label	Value Assigned	A	H	M
Not deprived	1 – 2 = 1	1/8 + 2/8 = 0.1875	465/1799 = 0.258	0.048
Slightly deprived	3 – 4 = 2	3/8 + 4/8 = 0.4375	482/1799 = 0.268	0.117
Moderately deprived	5 – 6 = 3	5/8 + 6/8 = 0.6875	273/1799 = 0.152	0.104
Highly deprived	7 = 4	7/8 = 0.875	325/1799 = 0.181	0.158
Absolutely deprived	8 = 5	8/8 = 1	254/1799 = 0.141	0.141

Source: The authors.

Finally, the ones deprived in all dimensions with 100 per cent deprivation, made up 14.1 per cent of the sample in this category. Their respective index values are shown in Table 6 (column 5). This does not present an encouraging picture as far as the deprivations in Punjab are concerned since 43.75 per cent of the sample was labeled as slightly deprived. If the definition is changed to a strict measure then the indices values will increase for the deprived bands. As many as 14 per cent of the totally deprived sample also does not encourage an analyst. These results corroborate the distressing picture presented by Salahuddin and Zaman (2012) and the OPHI MPI index for Pakistan (2011). According to one of the most elaborate conditional cash transfer programmes being run in Pakistan, the Benazir Income Support Program (BISP) Report 2011 states that the province of Punjab suffers from 21.08 per cent of poverty. This same report has elaborately listed districts according to proxy means testing and it reports that the districts of Southern Punjab show poverty ranging from 15.30 to 30.80 per cent.

Properties of Multidimensional Deprivation Index

The multidimensional poverty spectrum is a useful tool for the analysis of poverty in the following ways:

It is sensitive to any change in the spectrum band for individuals; therefore, helps to portray a realistic picture of poverty.

Similar to the Alkire–Foster measure, it also adjusts for the group size, resulting in international comparisons across various-sized countries.

It is a step-forward towards a closer look at the poverty picture in terms of the breakdown into different deprivation levels.

It can also depict a breakdown of poverty in terms of regions, gender, ethnic group, rural urban and other classifications.

Its properties of breakdown of different classes into various deprivation bands will be of great help to policymakers and governments of different countries for designing goal-targeted policies for poverty alleviation.

Conclusion

This study has argued that instead of classifying humans into poor and non-poor, poverty should be measured on the basis of its different dimensions. This argument has been operationalized by developing a multidimensional deprivation spectrum. It is a step forward from the Alkire–Foster multidimensional poverty index, where the methodology is similar being a dual cut-off method building on a triple stage deprivation spectrum using dimensions as the basic building blocks. It not only shows the level of deprivation within a dimension but also depicts the degree of intensity and spread across populations. The level of deprivation in Punjab depicted a distressing picture since; only 25 per cent of the respondent families were considered “not deprived”. These families were not entirely “not deprived” as they also showed 18.75 per cent intensity of deprivation. The rest of the 75 per cent of families in Punjab were deprived within the slightly deprived to absolutely deprived range. As many as 14 per cent were totally deprived in basic life dimensions such as education, living standard, air quality, water and sanitation, etc.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

References

Alkire

& Foster

(2007). Counting and multidimensional poverty measures (OPHI Working Paper No. 7). Oxford: Oxford University, Oxford Poverty & Human Development Initiative.

Alkire

& Seth

(2007). Measuring multidimensional poverty in India—A new proposal (OPHI Working Paper No. 15). Oxford: Oxford University, Oxford Poverty & Human Development Initiative.

Batchelder

A. B.

(1971). The economics of poverty. Retrieved from www.ebay.com/ctg/Economics-Poverty...Batchelder-1971...-/4594866

Besharov

D. J.

& Germanis

(2004). Reconsidering the federal poverty measure: Project description. University of Maryland.

Bos

A. E. R.

Pryor

J. B.

Reeder

G. D.

& Stutterheim

S. E.

(2013). Stigma: Advances in theory and research. Basic and Applied Social Psychology, 35(1), 1–9. Retrieved from http://dx.doi.org/10.1080/01973533.2012.746147

Bourguignon

(2002). From income to endowments: A difficult task of expanding income poverty paradigm. Paper prepared for the Conference on Conceptual Challenges in Poverty and Inequality Analysis, Cornell University, 16–17 April. Retrieved from https://ideas.repec.org/p/del/abcdef/2003-03.html

Chambers

(2006). What is poverty? Who asks? Who answers? Sussex, UK: Institute of Development Studies, International Poverty Centre Poverty in Focus. Retrieved from http://www.ipc-undp.org/pub/IPCPovertyInFocus9.pdf

Citro

C. F.

& Michael

R. T.

(1995). Measuring poverty: A new approach. Washington, DC: National Academy Press.

Foster

Greer

& Thorbecke

(1984). A class of decomposable poverty measures. Econometrica, 52(3), 761–765.

10.

Jencks

Mayer

S. E.

& Swingle

(2004). Can we fix the federal poverty measure so it provides reliable information about changes in children’s living conditions? (Working Paper). Harvard University.

11.

Rawls

(1999). The theory of justice. Harvard University Press.

12.

Salahuddin

& Zaman

(2012). Multidimensional poverty measurement in Pakistan: Time series trends. Pakistan Development Review, 51(4), Part II, 493–504.

13.

Sen

(1985). Commodities and capabilities. Amsterdam: Elsevier.

14.

Thorbecke

(2005). Multidimensional poverty: Conceptual and measurement issues. Paper prepared for The Many Dimensions of Poverty International Conference, UNDP International Poverty Centre, Brasilia, August 29–31.

15.

UNDP (2011). Sustainability and equity a better future for all. Human Development Report 2011. Retrieved from http://hdr.undp.org/en/content/human-development-report-2011

Multidimensional Deprivation Spectrum: A Step Forward from Alkire–Foster Methodology

Abstract

Keywords

Introduction

Methodology

Step-wise Approach

Empirical Testing

Multidimensional Deprivation Spectrum

Properties of Multidimensional Deprivation Index

Conclusion

Footnotes

Declaration of Conflicting Interests

Funding

References